What you would like to accomplish:

Please provide IPA transcripts or other pronunciation guides via the API. IBM Watson does this for nine languages. The Oxford English Dictionary does this for English, with multiple pronunciations for some words. Amazon AWS and Google Cloud don't provide pronunciation guides.

I have also made this request for Google Cloud Text-to-Speech.

How this might work:

"IPA" could be a language selection between Indonesian and Irish. Or IPA could be a third option in the new Romanization service, after Transliteration. https://cloud.google.com/translate/docs/languages#roman

Here's how IBM does it.

```js
const textToSpeech = new TextToSpeechV1({
authenticator: new IamAuthenticator({
apikey: '12345abcde',
}),
url: 'https://api.us-south.text-to-speech.watson.cloud.ibm.com/instances/abcde12345'
});

const getPronunciationParams = {
text: text,
format: format, // 'ipa' or 'ibm', the latter is designed for English speakers
voice: voice, // e.g., 'en-US_AllisonV3Voice', this sets the language
};

// get the IPA transcript from IBM Watson Text-to-Speech
function getIPA() {
return new Promise(function (resolve, reject) {
textToSpeech.getPronunciation(getPronunciationParams)
.then(function (ibmIpaObject) {
resolve(ibmIpaObject);
})
.catch(error => console.error(error));
});
}
```

If applicable, reasons why alternative solutions are not sufficient:

Romanization is a pronunciation guide and is perhaps easier for English speakers to understand than IPA transcripts. IPA transcriptions are the most accurate pronunciation guide.

IBM's nine languages are not sufficient for our needs. We need Arabic and Russian, which IBM doesn't provide. It would be easier for Google to add pronunciations guides than for IBM to add Arabic and Russian.

Issue 286713576

Links (2)

Issue metadata